In its fight to take down tech support scams worldwide, Microsoft relied on artificial intelligence (AI) to help it track down the scammers. In a Thursday blog post, the company detailed how its Digital Crimes Unit used the technology to figure out the methods employed by the scammers, and how they avoided getting caught in the past.

The scams in question typically occur when a pop-up window shows up, informing a user that they need to call for tech support to remove a virus or other form of malicious software. If the user calls the number listed, they are typically directed to someone trying to sell them services they don’t actually need, the post said.

The US Federal Trade Commission (FTC) announced an official crackdown on these scams with Operation Tech Trap in May 2017. According to a press release announcing the project, Microsoft was a partner in the operation.

SEE: Information Security Management Fundamentals (TechRepublic Academy)

Microsoft and the others involved in Operation Tech Trap had their work cut out for them. Chris White, a principal researcher at Microsoft who worked with the Digital Crimes Unit during the operation, said in the release: “These people are very clever.”

While the scams traditionally start out with an email or pop-up window, phone calls have been reported as well, especially in other countries. Of the tech scams reported to Microsoft from Germany, 85% started with a phone call, the post said.

Many of the users reporting the scams could describe what was happening, but who might have been behind the attack was still in question. “We had a bunch of customers who were reporting scams but didn’t know who scammed them,” Courtney Gregoire, assistant general counsel for the Digital Crimes Unit, said in the release.

After collecting a bunch of data that seemed like it could be part of a scam, Microsoft researchers used machine learning to look through it all, the post noted. The custom AI tools that Microsoft ran on its Azure cloud service learned from the data they were being fed what to look for. The team also used Microsoft’s computer vision API to pull additional relevant data from the content they collected, the release said.

Once they had the data collected and organized, the Microsoft team presented it to FTC officials to help with the investigation.

“This is a story of a practical application of bone fide machine learning to address an important problem,” White said in the release.

The 3 big takeaways for TechRepublic readers

  1. Microsoft used AI to combat worldwide tech support scams as part of the FTC’s Operation Tech Trap.
  2. Machine learning was used to sort through content that was collected to determine if it was part of the fraudulent scams.
  3. The team also used Microsoft’s computer vision API to collect data like phone numbers from content associated with the scam.